Commentary on Palmer, "Color, Consciousness and the Isomorphism Constraint"
Department of Linguistics and Philosophy
Cambridge, MA 02139
Abstract: Palmer's "subjectivity barrier" seems to be erected on a popular but highly suspect conception of visual experience, and his "color room" argument is invalid.
Palmer beautifully articulates a view that many philosophers and psychologists have found compelling: in attempting to understand the mind, scientists face an impenetrable "subjectivity barrier", behind which lies the "nature of [our] experiences themselves" (sect. 2.1, para. 11). And although Palmer sometimes sweetens the pill with such locutions as "scientists would never know with certainty" (sect. 2.1, para. 1), and "it may not be possible to be sure" (sect. 2.2, para. 3), it is perfectly clear that the conclusions of his arguments are--setting futuristic brainometers aside--that we (not just "behavioral scientists") have absolutely no reason to believe (not just no "objective knowledge") that others have experiences with the same "intrinsic qualities" as our own, or even that they have any experiences at all (sect. 2.5, para. 8). Thus, what might seem at first glance a sober essay curbing the pretensions of science to explain consciousness is in fact a radically skeptical manifesto.
According to Palmer, we can at best know about the relational structure of others' experiences--the similarities and differences they bear to each other. (For brevity's sake I shall ignore his further claim that we cannot know that others have experiences.) Skepticism follows because Palmer holds that fixing this relational structure does not fix the "sensory qualities" of experiences.
Palmer's main examples of a difference in sensory quality with no difference in relational structure are variants of the usual "spectrum inversion" case (Shoemaker 1981). No doubt some commentators will attack Palmer's argument by disputing that our color space is as symmetric as Palmer makes it out to be. But this response does not dig deep enough, because on Palmer's conception there may well be sensory qualities "alien" to an individual that could occupy the relational structure of his experiences.
A potentially more promising response disputes Palmer's crucial claim that sensory qualities can vary independently of relational structure. This response subdivides into two: the first agrees with Palmer that interpersonal comparisons of sensory qualities make sense (Clark 1993; Hilbert & Kalderon, in press), while the second denies that they do (Stalnaker, in press).
As I am unconvinced that the response just mentioned can be made to work, I shall try a different tack, and present a natural conception of visual experience on which it is hard to get Palmer's skeptical worries going.
Imagine seeing a red object--the proverbial ripe tomato. Focus your attention on the salient property of the tomato that it shares with cherries and strawberries. That property is redness, right? And of course red objects--like tomatoes and cherries--look more similar, in respect of color, to orange objects than they do to green objects. These similarity relations among colored objects induce corresponding similarity relations among our experiences of color, rather than the other way around. It's not true that red objects are more similar to orange objects than they are to green objects because our experiences of red objects are more similar to our experiences of orange objects than they are to our experiences of green objects. Rather, the explanation goes the other way: our experiences are similar because the objects of the experiences are similar. This is because when a person has a visual experience he is only aware of what visually appears to him, or of what his experience represents: as it might be, the presence of a red tomato. He is not aware (at least, not without further effort) of his experience. And similarly with the "nature of the experiences themselves": the perceived nature of the objects of perception (e.g. the redness of the tomato) explains the nature of perceptions of objects. If a person has a visual experience that represents the tomato as being red, then nothing else needs to be added to his experience for it to have the distinctive "sensory quality of redness" (sect. 2.1, para. 3) that Palmer thinks is hidden from scientific enquiry. Therefore, since there appears to be no special problem about knowing whether objects visually appear red to people, there's no special problem about knowing the nature of others' visual experiences, and thus there is no "subjectivity barrier".
Palmer may have been seduced by the following perennially appealing argument for thinking that something else needs to be added to a visual experience that represents something as red, in order for it to have the "sensory quality of redness". Imagine having a "red" afterimage. Perhaps unreflectively you would be inclined to call the distinctive property of the image "red", but surely the image can't really be red--red is a property of physical objects like tomatoes, not "mental objects" like images. So call the property of the image "R" instead. But obviously R is present when you see red objects, like ripe tomatoes. And since "mental objects" like images can have R, it doesn't seem likely that R can ever be a property of a physical object like a tomato. So when you see that a tomato is red, you are aware that the tomato is red and that some image-like thing has R. It is the presence of an R-image that gives your experiences of red objects (and certain afterimages) their distinctive "sensory quality"; similarly, the distinctive sensory quality of your experiences of green objects is due to the presence of a G-image. And now, of course, the question naturally arises whether others' experiences of red objects are in fact attended by G-images rather than R-images.
As tempting as this reasoning is, it is wrong. If R is to explain the sensory quality of your experience of a ripe tomato then it is not sufficient that the experience involve an image that has R: it must visually appear to you that the image has R (imagine that even though the image has R, it appears to have G: in that case you would have an experience with the sensory quality distinctive of your experiences of green objects). But now the alleged fact that the image has R isn't doing any explanatory work: the sensory quality of your experience is solely explained by the fact that it appears to you that the image has R, irrespective of whether the image has R. So the introduction of R was an idle wheel--redness would do the job just as well. Your afterimage experience was a kind of hallucination: it visually appeared to you that something was red (that's what gave your experience its distinctive sensory quality), but nothing in the scene before your eyes was red (that's why it was a kind of hallucination). Further, although it seemed to you that there was an image floating before your eyes, in fact there was no object--not even a mental one--there at all.
Finally, as has been pointed out numerous times (e.g. Copeland 1993), the Chinese room argument is fallacious. The conclusion concerns the system (it can't understand Chinese), but the premise concerns a part of the system (the man doesn't understand Chinese). The argument is an instance of "x isn't F, x is part of y, therefore y isn't F", and so is invalid. Thus Palmer's "color room" argument fails to show anything whatever about functionalism and experience.
Armstrong, D. M. (1968). A Materialist Theory of the Mind. London: Routledge.
Block, N. (1990). Inverted earth. Reprinted in N. Block, O. Flanagan, & G. Güzeldere (Eds.), The Nature of Consciousness. Cambridge, MA: MIT Press, 1997, 677-693.
Block, N. (1995). On a confusion about a function of consciousness. Behavioral and Brain Sciences, 18, 227-287. Reprinted with changes in N. Block, O. Flanagan, & G. Güzeldere (Eds.), The Nature of Consciousness. Cambridge, MA: MIT Press, 1997, 375-415.
Byrne, A., & Hilbert, D. R. (1997). Colors and reflectances. In A. Byrne & D. R. Hilbert (Eds.), Readings on Color, Volume 1: the philosophy of color. Cambridge, MA: MIT Press, 263-288.
Clark, A. (1993). Sensory Qualities. Oxford: Oxford University Press.
Copeland, B. J. (1993). The curious case of the Chinese gym. Synthese, 95, 173-186.
Dretske, F. (1995). Naturalizing the Mind. Cambridge, MA: MIT Press.
Frege, G. (1884/1950). The Foundations of Arithmetic (J. L. Austin, Trans.). Oxford: Blackwell.
Frege, G. (1918/1988). Thoughts. Reprinted in N. Salmon & S. Soames (Eds.), Propositions and Attitudes. Oxford: Oxford University Press, 33-55.
Harman, G. (1990). The intrinsic quality of experience. Reprinted in N. Block, O. Flanagan, & G. Güzeldere (eds.), The Nature of Consciousness. Cambridge, MA: MIT Press, 1997, 663-675.
Hilbert, D. R., & Kalderon, M. (in press). Color and the inverted spectrum. Vancouver Studies in Cognitive Science.
Lycan, W. G. (1996). Consciousness and Experience. Cambridge, MA: MIT Press.
Shoemaker, S. (1981). The inverted spectrum. Reprinted in N. Block, O. Flanagan, & G. Güzeldere (eds.), The Nature of Consciousness. Cambridge, MA: MIT Press, 1997, 643-675.
Stalnaker, R. (in press). Comparing qualia across persons. Philosophical Topics.
Tye, M. (1995). Ten Problems of Consciousness. Cambridge, MA: MIT Press
 Palmer cites Wittgenstein as a supporter, but I think the reverse is true: the view Palmer holds is one that Wittgenstein argued against. Frege--the inventor of modern logic and one of the founders of analytical philosophy--is a much better candidate (see especially 1918/1988, and also 1884/1980, §26, where Frege uses the dual system analogy of the target article (sect. 2.3 paras. 11-12)).
 Stalnaker draws a helpful analogy with Von Neumann-Morgenstern utility theory, which assigns utility scales to people that can only be compared intrapersonally, not interpersonally.
 See Armstrong 1968; Harman 1990; Lycan 1996; Dretske 1995; Tye 1995; Byrne & Hilbert 1997. There are important dissenters, in particular Block 1990 and 1995.
 Searle's response is to let the man perform all the symbolic manipulations in his head, but this appeals to the inference pattern "x isn't F, y is part of x, therefore y isn't F", which is also invalid.